#calibration
Calibration: 4 AI articles covering calibration news, analysis, and research
Articles
Abstention Is Not Threshold Compliance: Stated Scoring Rules inβ9
Open language models often abstain due to prompt format, not stated thresholds; a study of 12 checkpoints finds threshold compliance largely fails across scorin...
ThinkJev: A Hybrid Fast-and-Slow Framework for Calibrated Typedβ9
ThinkJev is a training-free fast-and-slow framework that escalates to reasoning only when needed, yielding calibrated typed decisions with controlled error from...
HakemBench: A Turkish Benchmark of Typed Decisions for LLM Evaluationβ8
HakemBench introduces a Turkish benchmark of 2,346 items for evaluating LLM probabilistic decisions across seven tracks using calibration-aware metrics and boot...
On-Device Named-Entity Recognition: A Deployability Study ofβ9
A deployability study of on-device named-entity recognition across nine systems, evaluating accuracy, latency, output validity, and the trustworthiness of model...
